The future arrives for me in the form of a rectangle on four wheels, about half the size of a conventional school bus, with large windows all around. When the side door opens, an impressively mustachioed safety attendant named Buddy welcomes me aboard. Standing at a small command terminal, Buddy cues up the route and disengages the emergency brake, and we peel out of the parking spot at a cool 11 miles per hour. This is no David Hasselhoff-ian Knight Rider driving experience—then again, KITT was science fiction. I put my sunglasses on and kick back as Olli comes to a gradual stop at a four-way intersection and then navigates a left turn seamlessly.
This is my inaugural ride on the Olli shuttle bus, and my first-ever ride in an autonomous vehicle, seated comfortably on a 3D-printed seat with a belt fastened across my lap, cruising along at a speed attainable by a skilled tricycling kindergartener. With Buddy. And his garage-broom-bristle ’stache.
On a Wednesday afternoon in January, I’m in National Harbor, the Maryland-based sales and demonstration facility operated by Local Motors. The Arizona-based automaker is known for crowdsourcing vehicle design and then creating those vehicles using advanced manufacturing techniques. (It was Local Motors that created the Strati, the first fully 3D-printed car. The Olli shuttle I rode on was designed by Colombian-born Edgar Sarmiento; roughly 80 percent of the thing is 3D-printed.) Chances are you’ve heard of Olli, especially if you live in the Washington, D.C., area. For the last several years, Local Motors has teased a wider rollout of its electric autonomous shuttle bus in the D.C. area.
In 2019, Olli seems to have finally found its wheels. The ride I took in January marks phase one of Local Motors’ pilot project in National Harbor, the waterfront dining and entertainment district just south of the nation’s capital on the Maryland side of the Potomac River. This year, Local Motors is offering autonomous rides on the Olli, along fixed routes around the neighborhood’s main shopping and eating areas. Later this year, Local Motors will increase the number of buses operating along these routes from one to three, part of a pilot project to prove not only that a higher number of self-driving shuttles can operate safely, but also that the demand for these shuttles exists in the first place.
“We believe that Olli isn’t a replacement of one specific type of transportation, but an enhancement of a multimodal transportation system,” says David Woessner, executive vice president of corporate development and regulatory affairs for parent company Local Motors Industries.
Woessner envisions a transit system where connections between more conventional bus, rail, and ride-hailing networks are served by self-driving shuttles rather than individual cars. Need to get to a building on campus? Hop on the Olli. Going from the airport to the nearest Metro station? Olli. Staying at a hotel in National Harbor and want to try your luck on the roulette wheel at the MGM Grand a short drive away? Take the Olli.
Most of the attention on the autonomous-driving industry falls on the major self-driving car companies. Yet the storyline of the last few years is one of tempered expectations. Last July at the National Governors Association meeting, John Krafcik, the CEO of Waymo, got on stage and told those assembled that it will be “longer than you think” for self-driving cars to arrive on America’s roads.
“There are no autonomous systems available, zero on the road today,” Krafcik said.
He must not have been thinking about buses.
In just the last two years, a movement toward driverless, electric shuttle services in America’s cities has quietly bubbled up. In Detroit, Las Vegas, Columbus, even Lincoln, Nebraska, autonomous mini-buses have completed successful shuttle trials or are currently ferrying passengers.
An autonomous shuttle that hit the road in San Ramon in March 2018 was the first vehicle to operate on California’s roads without a driver behind the steering wheel. Three months later, five autonomous shuttles operated by Michigan-based startup May Mobility began carrying Bedrock and Quicken Loan employees in the Detroit downtown core.
Perhaps the biggest pilot of all is a state- and privately funded autonomous shuttle program happening now in Columbus, Ohio. Three AV shuttles from May Mobility follow a 2.7-mile route in downtown Columbus, and more than 3,300 riders have climbed aboard since the shuttles began running early last year. A second, federally funded route in an entirely residential neighborhood will launch in November.
Admittedly, these shuttle projects are small, running in confined geographical areas on downtown roads and not highways, and are restricted in speed. The shuttles in Columbus don’t travel any faster than 25 miles an hour. Evidence regarding ridership, safety, and operating efficiency is scarce so far. Still, companies like Local Motors see an opening for small AV shuttle buses.
I couldn’t help but notice, the day I rode in an Olli, that if you were to shrink the thing down to palm size, it would make for an adorable children’s toy.
In several major urban areas, including Los Angeles, Miami, Chicago, Philadelphia, and Washington, D.C., public transportation ridership was lower in 2016 than it was in 2006, according to data from the American Public Transportation Association. Much of the decrease is concentrated on bus lines: In 1990, buses made up almost two-thirds of all public transit trips in America. By 2014, that figure was under 50 percent.
Instead of turning cities off to public transit, these declining ridership numbers are only spurring more of them to undertake dramatic redesigns of their bus systems—and Olli and other autonomous shuttles, the thinking goes, can act as supplemental pieces to existing systems of public transit.
“We see a lot of cities doing bus redesigns right now, because their current model of fixed-route, fixed-schedule is not working very well,” says Ashley Z. Hand, co-founder of smart city advisory practice CityFi and the former transportation technology strategist for the City of Los Angeles. “An AV shuttle might be very complementary to redesigned systems by getting people to those connections in the first place.”
If the autonomous shuttle industry wields any advantage over the more conventional self-driving car companies, it’s consumers’ lower expectations. No one’s looking for a steering-wheel-less shuttle bus to be scuttling along the right-bound lane on the freeway doing 65 mph in the same way that folks are anticipating self-driving cars will do one day. In that sense, Waymo’s CEO was dead-on in saying no autonomous systems are available on today’s public roads.
Per the updated guidelines released last summer by the Society of Automotive Engineers, most of the autonomous buses operating could be termed Level 3 autonomous systems, where a human will be expected to intervene in some traffic scenarios. At best, AV shuttles are only Level 4, in which no human intervention is expected but the shuttle requires a protected environment where the vehicle is guaranteed not to encounter any difficulty: They travel on mapped, pre-planned routes, meaning there are no surprises. To some degree, the decisions they’re making on the road have already been made for them.
“If a tree falls and blocks the road and the automated system can’t see around it, we’re not at the level where the automated system can handle that,” says Aaron Foster, a solutions engineer at the Michigan-based production plant for Navya. The French company, whose eight-seat shuttles are deemed Level 3, sells to countries worldwide, and built out a U.S. base of operations by making its shuttles available for testing and several pilot projects, like the one in Las Vegas.
Considered one of the longer shuttle pilot projects, the Las Vegas test kicked off in November 2017 and used a Navya shuttle that traveled along a 0.6-mile loop in the city’s Fremont East Entertainment District. The time of day when the shuttle ran was circumscribed: 11 in the morning to 7 at night, Tuesday through Sunday. The shuttle’s loop was merely a back-and-forth path along a closed eastbound lane. On top of the sensor technology already available on the shuttles, six short-range radios allowed the shuttle to wirelessly communicate with traffic signals along the route and know when it was time to go, slow, or stop.
The slower speed at which shuttle deployments take shape is a key advantage, according to Joanna Wadsworth, project manager for the Las Vegas shuttle pilot. The city is already trying to determine how driverless vehicles might change up traffic patterns or pedestrians’ expectations; a small shuttle whose ultimate use will more than likely be running specific, low-speed routes in a downtown area just makes that easier to figure out, especially since residents and visitors don’t expect a ride on an AV shuttle in the first place.
“It performed really well, and really provided a hands-on experience for our city team to understand what technology we might face and will need to provide for vehicles in the future,” she says.
All in all, 32,000 passengers took a trip on the Vegas shuttle by the time the project concluded last October, and the bus was involved in only one crash—the result of a delivery truck backing into it on the shuttle’s first day on the road. Wadsworth says the city is already looking toward a future deployment, one where AV shuttle buses are providing connections within an existing seven-stop downtown loop served by a conventional bus that runs seven days a week.
People need to see autonomous public transit, and see that it gets them where they need to go just as efficiently, in order for them to choose that over their own car, a ride provided by Uber or Lyft, or, someday in the future, their own driverless car.
For companies working on small autonomous shuttles, that’s the killer app: filling in the first-mile, last-mile gaps within broader public transportation networks.
Since beginning operations in Detroit in June 2018, May Mobility has conducted 50,000 rides. Originally, in the area where May Mobility’s shuttles now run, there was a 30-passenger diesel bus that operated in a one-mile loop downtown taking people to and from parking garages and office buildings. Now that bus has been replaced by three six-passenger electric shuttles.
“It’s not just about getting AVs on the road; it’s about solving transportation needs,” says Alisyn Malek, COO of May Mobility. “Are we getting people to work? Are we reducing congestion?”
In Finland, Switzerland, and France, where autonomous shuttles are entering their third and even fourth years of operation, that sort of dynamic is already playing out, says Navya’s Foster. Although the adoption of public transportation in Europe is higher than in the U.S., he says the same sorts of conditions that make AV shuttles practical in European cities—highly congested roads, the struggles of urban parking, and decreasing ridership on traditional public transit—are present in American cities, too.
“There’s a niche here for microtransit shuttles,” Jeremy Mulder, vice president at May Mobility, told Curbed last summer. “The technology today can support slow-speed, 15-mile-per-hour, reliable travel in the urban core.”
Yet the potential for small AV shuttles to reshape urban transportation networks rests on a few still-unrealized assumptions.
A system like the one in Detroit solves one problem: By taking people between parking garages and office buildings, it addresses the lack of street parking often found in downtown cores. But it still means people need to drive their cars to those parking garages in the first place. AV shuttles might eventually circumvent this problem given their smaller axle-print—there’s a greater opportunity for more-maneuverable smaller shuttles with fewer seats to handle dynamic, on-demand trips that can move people closer to specific destinations.
But that same smaller axle-print might cause a second problem when considering how to rejigger downtown transit turmoil.
“Lots of little buses can offer as much capacity as a big bus, but the little buses take more space per person. In a dense city, we still need to prefer the big buses, which use space more efficiently,” says public transit consultant Jarrett Walker. “So big benefits will arise only when we have a big driverless bus that can handle a sufficient range of conditions.”
As cities become denser, more crowded, and therefore more congested, says transportation planner Jeffrey Tumlin, the more natural shift to autonomous technology will come in the form of high-capacity vehicles.
“Cities have a much higher density of travel demand,” he told Curbed in 2017. “Transit agencies need to be the first adopters of AV technology, or public transit dies in America.”
Welcome to the third challenge. Whether it’s small shuttles steering various routes in low-density areas or big buses navigating crowded city streets without careening into a storefront, the technology will have to be able to keep up. And improvements in autonomous vehicle technology could actually lower demand for autonomous shuttles.
The rise of Uber and Lyft across the U.S., instead of mitigating vehicle congestion, actually made it worse. People take more car trips because taking those car trips is now easier. Why walk 20 minutes to a downtown market when a rideshare is a phone tap away? From the data Hand has studied, improvement in AV technology seems to indicate an increase in total miles traveled by vehicles, not the abandonment of cars in favor of autonomous public transit.
“Imagine never having to look for parking because you have your robot car circle the block while you run your errand. This can make congestion in cities so much worse,” she says.
People need to see autonomous public transit, and see that it gets them where they need to go just as efficiently, in order for them to choose that over their own car, a ride provided by Uber or Lyft, or, someday in the future, their own driverless car. Getting people to trust autonomous transit still presents a slight challenge, although pilots like the one that ran in Las Vegas help ease curious riders into the experience. (So does a twee design: I couldn’t help but notice, the day I rode in an Olli, that if you were to shrink the thing down to palm size, it would make for an adorable children’s toy.) But shorter trips, like heading to lunch or running a quick errand, could be well served by an autonomous shuttle.
“What we ultimately recommended in Los Angeles is to deploy autonomy as quickly as possible in public transit now,” Hand says. “The trick is how do we make sure that people choose the right mode for the right trip.”
In addition to its pilot in National Harbor, Local Motors recently launched another at Sacramento State, where two Olli shuttles drive students and faculty around campus. Another pilot is running in Phoenix, not far from Local Motors’ headquarters in Chandler. They represent the first two “fleet challenges” the company has set up in the U.S., a kind of reverse request-for-proposal process where municipal governments and organizations from selected cities submit potential use cases for AV shuttles, which Local Motors then fulfills using its Olli buses.
“We select the most valid use case we think the technology is ready to tackle, and we’re going to place vehicles in those locations for up to three months,” says Woessner.
By and large, the AV shuttle industry thinks there’s enough ridership data from various pilot programs to justify continued expansion of small autonomous buses on public roads. As Walker contends, small AV shuttles will be useful in niche cases and in low-density areas, places where the need for several of them on the road in order to meet ridership demands won’t lead to further road congestion.
Yet two key hurdles, one technological and the other regulatory, remain.
“With these shuttles, there’s an intense amount of support you need around a deployment,” says Carrie Morton, deputy director of the Mcity automated driving test facility at the University of Michigan, where Navya shuttles also operate on a closed loop.
The mile-long test route comprises roads that run through a research complex at the university. While there are pedestrians and other cars on the route, the situation is quite controlled: The shuttles don’t drive any faster than 12 miles an hour, and they can’t currently drive around obstacles. According to Morton, the on-board safety conductor takes manual control to steer the bus around vehicles parked in the road.
“These rely on high-definition maps, and you need communication with the route and connectivity to infrastructure,” she says. Right now, Mcity’s shuttle test program is on hiatus: Construction along the route makes it impossible for the AV buses to run. That’s the sort of curveball routing challenge small AV shuttles are not yet ready to confront.
“The challenge we have with AV shuttles is, because of the technological limitations we have right now, they’re not quite ready for dynamic routing,” Hand says. “If you have a fixed route, you can anticipate the intersections and conflicts with transit and with pedestrians.”
After its splashy preview of the Olli shuttle in 2016, the technological problems Local Motors encountered sent it back to the drawing board.
“To be quite candid, the technology has improved greatly over the last three years,” Woessner says. “We went and worked pretty much all of 2017 to find what we thought was the best-in-class autonomous capability for the use cases and applications that we had.”
A Maryland-based company, Robotic Research, now produces the key autonomous software Olli needs in order to do things like stop at the four-way intersections that are on its route in National Harbor. Research deployments—one at Local Motors HQ in Arizona and another two at the University of Buffalo and in Copenhagen, Denmark—serve as proving grounds, similar to Mcity, for testing the upgraded AV technology. That’s what enabled Local Motors to relaunch the Olli in the first half of 2018.
But for AV shuttles to make a meaningful dent in the decline of public transit nationwide, they’ll need to get the safety attendants out of the vehicles entirely.
“Transit operating cost is mostly labor, so driverless, if it were ever achieved, would transform public transit by radically increasing frequency, the single most important driver of transit ridership,” Walker says. “But you have to have Level 4 or better. If there’s a paid human in every vehicle, you don’t get any savings even if they’re not driving.”
When it comes to the regulatory landscape, the companies running small AV shuttles have largely relied on local governments and municipal agencies to pave the way. Because something like the Olli bus is non-compliant by Federal Motor Vehicle Safety Standards—no steering wheel, no brake pedal, no rear or side mirrors—it requires a federal exemption issued by the National Highway Traffic Safety Administration to deploy widely on roads across America. Those exemptions usually take months to receive, though the safety rules for fully driverless vehicles could be changing soon: The Department of Transportation plans to rewrite safety rules for cars, as it announced last fall. To run its pilot in National Harbor, Local Motors instead went directly to the Maryland Department of Transportation to craft a special permit.
“Long-term, in order for us to be commercially viable, the overall regulatory framework needs to change,” says Woessner. The company’s goal, he says, is to start selling commercially by 2020. For that reason, a third fleet challenge is also up and running in Adelaide, Australia, where the regulatory framework is much less burdensome, and will go for six months instead of the usual three.
AV shuttle companies continue to press on, despite the uncertain regulatory future on the national level. May Mobility will launch new shuttle routes in Grand Rapids, Michigan, and Providence, Rhode Island, before the end of 2019; the Providence route will be a five-mile loop connecting an Amtrak line to commuter rail service. And Local Motors is currently looking over proposals submitted for its fourth fleet challenge, this one in the Washington, D.C., metro area. By early May, Woessner says another Olli shuttle route will be operating.
“At the end of the day we’re manufacturers, and we want to sell fleets of vehicles,” he says. “I’m not a transportation or transit planner, but Olli has a set of use cases that other modes don’t.”
Andrew Zaleski, a journalist based near Washington, D.C., writes frequently about business, science, and technology.
Correction: An earlier version of this article said the Columbus, Ohio, autonomous shuttle pilot was federally funded. The Columbus route is privately and state-funded. An upcoming second route in the city will be federally funded.